BACKGROUND: The accuracy of the World Health Organization method of estimating malaria parasite density from thick blood smears by assuming a white blood cell (WBC) count of 8,000/muL has been questioned in several studies. Since epidemiological investigations, anti-malarial efficacy trials and routine laboratory reporting in Papua New Guinea (PNG) have all relied on this approach, its validity was assessed as part of a trial of artemisinin-based combination therapy, which included blood smear microscopy and automated measurement of leucocyte densities on Days 0, 3 and 7. RESULTS: 168 children with uncomplicated malaria (median (inter-quartile range) age 44 (39-47) months) were enrolled, 80.3% with Plasmodium falciparum monoinfection, 14.9% with Plasmodium vivax monoinfection, and 4.8% with mixed P. falciparum/P. vivax infection. All responded to allocated therapy and none had a malaria-positive slide on Day 3. Consistent with a median baseline WBC density of 7.3 (6.5-7.8) x 10(9)/L, there was no significant difference in baseline parasite density between the two methods regardless of Plasmodium species. Bland Altman plots showed that, for both species, the mean difference between paired parasite densities calculated from assumed and measured WBC densities was close to zero. At parasite densities <10,000/muL by measured WBC, almost all between-method differences were within the 95% limits of agreement. Above this range, there was increasing scatter but no systematic bias. CONCLUSIONS: Diagnostic thresholds and parasite clearance assessment in most PNG children with uncomplicated malaria are relatively robust, but accurate estimates of a higher parasitaemia, as a prognostic index, requires formal WBC measurement.